How Outlaw Sake Used AI to Validate Marketing Ideas In Days Vs. Months

Generative AI is quickly becoming a new norm in marketing, with teams using the technology to experiment with creative concepts and accelerate campaign production. In fact, Salesforce research finds that more than half of marketers today are using gen AI for processes including basic content creation (76%), writing copy (76%), and generating images (62%).

It’s evident that marketers can leverage gen AI to quickly produce branding and ad prototypes. But how can they validate whether this technology is actually fueling productive brand development and driving lift? 

Outlaw Sake, a sake brand under development that’s aiming to break into the hard seltzer market, dramatically sped up its go-to-market strategy by using gen AI to conceptualize and create its first ads. Outlaw Sake followed up by rapidly pre-testing those concepts through Swayable’s randomized control trial (RCT) methodology. Insights from our rapid lift measurement tests helped the brand validate that some of its humorous, irreverent, and self-aware ad concepts could positively sway target audiences, while others were ineffective at driving positive impact.  

The opportunity: quickly develop a brand that disrupts the hard seltzer market 

Hard seltzer is booming: Precedence Research reports that the U.S. hard seltzer market size is slated to increase from $8.7 billion in 2025 to nearly $40 billion by 2034, at a CAGR of 15.5%. This growth corresponds to the trend of American drinkers seeking inexpensive, convenient, low-ABV (alcohol by volume) alternatives, driven by popular canned beverages such as White Claw and High Noon. 

Outlaw Sake co-founders Felix Yip and Darren Himebrook, partners at marketing agency Obsssn.co, saw an opportunity to jump on this trend by introducing a new hard rice seltzer that fit the profile and challenged the status quo of traditional Japanese sake. Their goal was to design a mainstream sake for all Americans, positioning the product as a casual everyday drink rather than a beverage that’s only suitable for sushi dinners. 

“We know that Americans are choosing lighter alcoholic beverages, and sparkling sake fits that profile . But the challenge is that sake is bound by category conventions: it can come across as too artisanal and too tied to Japanese traditions,” said Yip. “Why should you only drink sake with sushi?”

Enter Outlaw Sake: a brand whose values are defiant, unapologetic, individualistic, and American. 

The challenge: validating marketing ideas with the help of gen AI

In an effort to speed up creative development, Outlaw Sake opted to use AI to generate multiple video ads designed to introduce the brand to consumers.

All of the ads shared a common theme: challenging the status quo of traditional sake. But the ads communicated this theme through different messaging and creative that ranged from humorous to irreverent to outrageous. The ads featured amusing narration and concepts encapsulating the brand’s identity, including: 

“Sa-keh,” where a man cheekily explaining the difference between Outlaw’s product and traditional sake.

“I’m an AI,” which is transparent about the fact that the ad was entirely AI-generated.

“Alligator,” which promotes the product through a talking alligator.

“Genius,” which depicts people drinking sake at everyday functions.

Other ad treatments included people giving the middle finger (simulating the brand logo) and drinking sake with spaghetti—a callback to a viral AI-generated video of Will Smith eating spaghetti. 

The approach: rapidly pre-testing ideas through Swayable’s AI-powered platform

Outlaw Sake rapidly pre-tested these videos with consumers through Swayable’s platform. Using RCT methodology, we surveyed more than 7,000 people to quickly generate results about the creative’s impact on lift metrics including favorability, consideration, purchase intent, and authenticity. 

Within our dashboard, Outlaw Sake’s team was able to segment the results to focus on their target audience of hard seltzer and sake drinkers. AI was also a core part of our testing process: machine learning powers our platform’s population modeling, rapid insights extraction, and analysis automation of consumer responses.

After completing the experiment, one of the brand team’s main takeaways—apart from what the actual lift metrics told them—was how quickly they were able to produce, test, and prove the impact of the content using AI. It took only three days for Outlaw Sake to create the treatments versus the three weeks it could take to manually schedule and shoot videos. And it took only 24 hours to pre-test them using Swayable, versus waiting an entire month to complete a traditional brand lift study. 

“By using AI to generate creative and pre-test your ads, you can have a brand new idea validated or disvalidated within a four-week period instead of a four-month period,” said Yip. 

The outcome: metrics point to a distinct correlation between authenticity and lift

By deploying rapid pre-testing through Swayable, Outlaw Sake uncovered some illuminating insights about how consumers would react to its AI-generated content. Among sake and seltzer drinkers, in particular, nearly all of the pre-tested ads drove positive lift in favorability, consideration, and purchase intent. 

The three ads that generated the most impact in lift were the treatments that leaned into funny, irreverent concepts encapsulating the brand’s identity: “Sa-keh,” “I’m an AI,” and “Alligator.” These ads generated a 4.6, 2.8, and 2.4-point increase in lift, respectively, over sake drinkers who didn’t see the ad. Click through the dashboard below for a full snapshot of Outlaw Sake’s results: 

Not only did those three ads generate the most lift—sake drinker respondents deemed them the most authentic, demonstrating a strong correlation between lift and authenticity. Swayable’s qualitative analysis of comments also supported the idea that the more Outlaw Sake leaned into humor or self-awareness around AI, the more likely respondents were to perceive the brand as authentic—and the less likely they were to call out AI usage in a negative light (or comment about the technology at all). 

 “In terms of testing and category metrics, these results speak strongly to a concept, product, and brand that might actually compete well on the shelf,” said Himebrook. 

Additionally, the pre-testing results supported a classic Swayable observation: engagement metrics and upper-funnel lift metrics aren’t necessarily connected. Outlaw Sake’s ads that received higher engagement on Instagram and TikTok didn’t have the highest increases in lift, supporting the idea that content generating the most views on social media isn’t necessarily the most persuasive. 

While the metrics and insights pointed Outlaw Sake in the right direction for future marketing content to produce, the findings also uncovered a key insight about using AI: When used strategically and intentionally, machine learning can dramatically speed up the marketing creation and testing process, and deliver insights that are useful for decision making—even if the concepts aren’t fully fleshed out. 

“The ads you test don’t have to be polished—there can be enough there conceptually to help marketers make better decisions,” added Himebrook. “You need to get comfortable with experimenting and trial and error, especially when using AI-generated content. This process will help teams become more agile."

Our final takeaways from Outlaw Sake’s AI experiment: 

• Marketers can use AI to quickly validate or refute ad ideas through rapid RCT pre-testing. The ability to generate lift insights before spending money on consumer-facing concepts promotes a more proactive, agile marketing program. 
• More often than not, consumers will be able to identify an AI-generated ad. It’s vital to be transparent about using the technology. Leaning into self-awareness around AI boosted Outlaw Sake’s authenticity scores among target demos.
• Ads that generate the most social engagement aren’t always the most persuasive—meaning they may not be the best options for driving brand love or purchase intent. Rapidly pre-testing messages through Swayable’s platform helps brands avoid the engagement trap and prioritize content that actually sways consumers. 

Do you want to quickly test and confirm whether your AI-generated marketing ideas can drive lift? Book a demo with Swayable today.

Methodology

This monadic RCT survey experiment was conducted on Swayable over 26 hours between March 8 and March 9 with responses from 7,716 U.S. consumers aged 21+. Questions were asked of a general population sample aged 21+ using Swayable’s proprietary online platform.

The sample frame is U.S. smartphone users in all 50 states with active internet connections who are users of popular mobile and web apps that make up Swayable’s network of respondent partners. Respondents are solicited from partner apps with non-monetary reward offers for their participation. Respondents were recruited with an approximately even ratio of men and women imposed via separate quotas for each. This is a “non-probability sample” (in the conventional terminology of public opinion research, although this team believes this concept is not meaningful since truly random sampling of the population is not possible via any methodology). To correct for over/under-sampling, all samples are post-stratified to the general U.S. population using cross-tabulations accounting for factors including age, ethnicity, gender, educational attainment, and geography, based on the latest available data from the U.S. Census Bureau and Swayable’s proprietary population modeling. Margins of error quoted are based on response distribution statistics and sample sizes and are calculated independently for each result.

This research was conducted and self-funded by Swayable.